Second Order Effects Framework

A framework for understanding how incentive systems can lead to unintended consequences and second-order effects in business and organizations.

Core Concept of Second Order Effects

• First order effects are the immediate, intended outcomes of an incentive
• Second order effects are the unintended consequences that follow
• Most incentive systems successfully achieve first order goals but fail due to second order problems

Real World Examples

• Private Prisons

  • First Order: Paid based on filled beds
  • Second Order: Creates incentive to keep people imprisoned rather than rehabilitated
  • Proposed Solution: Pay partially for filled beds + bonus if inmates don't return within 4 years

• Sales Teams

  • First Order: Commission drives more sales
  • Second Order: Sales people close "dumb deals" with poor fit customers
  • Challenge: Hard to track quality metrics alongside pure sales numbers

• The Cobra Effect (Historical Example)

  • First Order: Bounty offered for dead cobras to reduce population
  • Second Order: People started breeding cobras to collect bounties
  • Result: Ultimately increased cobra population

Key Principles for Better Incentive Design

• Consider loss aversion

  • People feel losses more strongly than equivalent gains
  • Example: Giving 50% bonus upfront with clawback provision is more effective than end-of-period bonus

• Overattainment incentives

  • Set bonus thresholds at 105% of goals rather than 90%
  • Research shows people tend to achieve higher targets when structured this way

• Track multiple metrics

  • Need systems to measure both quantity and quality
  • Challenge: Tracking systems can become complex and expensive to implement

Implementation Challenges

• Requires bold experimentation with new incentive structures
• Tracking results/outcomes can be technically difficult
• Need to predict potential gaming of the system
• Must balance multiple competing interests
• Often involves significant operational complexity

The key is recognizing that while first order effects are usually predictable, second order effects require deeper analysis and creative solutions to avoid unintended negative outcomes.